Article: An intelligent recommendation method for multimodal ideological and political education resources integrating with dynamic portrait modelling Journal: International Journal of Business Intelligence and Data Mining (IJBIDM) 2026 Vol.28 No.10 pp.183 - 205 Abstract: Aiming at the limitations of traditional recommendation methods in accuracy, recall, and user experience, a new intelligent recommendation method for multimodal ideological and political education resources integrating with dynamic portrait modelling is proposed. A user profile system is constructed through multi-source data collection, employing spectral clustering and analytic hierarchy process to achieve fine-grained modelling of user attributes. A dynamic interest label update mechanism based on TF-IDF and a time decay function is introduced to track the temporal evolution of user interests. A hybrid recommendation model, LightGCNWithGAT, which combines graph attention mechanism with lightweight graph convolution, is designed to recommend multimodal ideological and political education resources that align more closely with user interests. Experimental findings show that the proposed approach attains a maximum accuracy of 97.89%, along with a recall as high as 98.03% and a user satisfaction level reaching 98.12%. Inderscience Publishers - linking academia, business and industry through research

Title: An intelligent recommendation method for multimodal ideological and political education resources integrating with dynamic portrait modelling

Authors: Xiaojin Wang

Addresses: School of Marxism, Hunan Information Vocational and Technical College, Hunan, 410200, China

Abstract: Aiming at the limitations of traditional recommendation methods in accuracy, recall, and user experience, a new intelligent recommendation method for multimodal ideological and political education resources integrating with dynamic portrait modelling is proposed. A user profile system is constructed through multi-source data collection, employing spectral clustering and analytic hierarchy process to achieve fine-grained modelling of user attributes. A dynamic interest label update mechanism based on TF-IDF and a time decay function is introduced to track the temporal evolution of user interests. A hybrid recommendation model, LightGCNWithGAT, which combines graph attention mechanism with lightweight graph convolution, is designed to recommend multimodal ideological and political education resources that align more closely with user interests. Experimental findings show that the proposed approach attains a maximum accuracy of 97.89%, along with a recall as high as 98.03% and a user satisfaction level reaching 98.12%.

Keywords: dynamic portrait modelling; multimodal; ideological and political education resources; intelligent recommendation; dynamic interest label update mechanism; LightGCNWithGAT.

DOI: 10.1504/IJBIDM.2026.156201

International Journal of Business Intelligence and Data Mining, 2026 Vol.28 No.10, pp.183 - 205

Received: 25 Feb 2026
Accepted: 09 Jul 2026
Published online: 07 Sep 2026 *

Title: An intelligent recommendation method for multimodal ideological and political education resources integrating with dynamic portrait modelling

Keep up-to-date